LOW DISTORTION TRANSIMPEDANCE AMPLIFIER USING DISTORTION CANCELLATION TECHNIQUE
BACKGROUND OF THE INVENTION
This invention is related to optical receivers and more particularly to a front
end amplifier for use in such receivers.
Optical receivers are typically utilized in communication systems for receiving
optical signals and transforming them into electrical signals. The electrical signals
are processed to achieve a desired electrical signal output level for further
transmission within the communication system. An example of such a
communication system is a CATV network.
A CATV network typically consists of a downstream path extending from a
service provider location known as a head end to various nodes through taps down
to settop terminals located at subscriber locations. The upstream path typically
extends from the settop terminals back to the head end. These communication
systems can be designed to be partially optical and partially electrical. For example,
communications between the head end and the nodes in both the upstream and
downstream directions can be accomplished utilizing optical signals while
communications in both directions between the nodes and settop boxes can be
electrical. Such a system requires optical receivers to be located at both the nodes
and headends.
Optical receivers are necessary at each node for receiving optical signals in
the downstream path and transmitting electrical signals to the settop terminals.
Likewise,
optical receivers are necessary at the head end to receive optical communications
from the nodes traveling along the upstream path.
Since CATV systems are becoming increasingly bidirectional, bandwidth
requirements along the upstream path are also increasing. The increased bandwidth
requirements are attributable to the need for upstream communications associated
with Internet access, fax capabilities, pay per view and other upstream information
transfer.
Transimpedance amplifiers have been widely used in optical digital receivers.
The advantages of using transimpedance amplifiers are their simplicity and low
Equivalent Input Noise Current (EINC) when the feedback resistance is large.
However, the large second order distortion produced by the transimpedance design
prevents it from being used in analog applications.
Matsuoka et al. showed in Electronics Letters Volume 18, No. 9 of April 29,
1982 that, using a GaAs MESFET with a silicon PIN photo diode in a
transimpedance amplifier, the second order distortions can be much lower than the
distortions created by transimpedance amplifiers using bipolar transistors. The
results presented were limited to low optical input powers (about 30 micro watts).
In applications where the maximum optical power approaches 1 mW, a new method
which produces low second order distortions is needed.
SUMMARY
The invention addresses the above mentioned problem by providing a method
and circuit for reducing second order distortion levels in a transimpedance amplifier.
A transimpedance amplifier is provided having a first input stage, a voltage amplifier
stage and an emitter follower stage. Distortion cancellation is accomplished by
determining the distortion generated by the first input stage and by controlling the
output impedance and bias current of the emitter follower stage. The second order
distortion generated by the emitter follower stage can be made equal to the
distortions produced by the first stage but with a 180° phase shift. Thus all the
second order distortions can be canceled. The advantages of transimpedance
amplifiers in simplicity and low EINC are still preserved.
BRIEF DESCRIPTION OF THE DRAWINGS
The invention will now be described by way of example with reference to the
accompanying figures of which:
Figure 1 is a block diagram of the transimpedance amplifier according to the
present invention.
Figure 2 is an exemplary schematic of an implementation of the block
diagram of Figure 1.
Figure 3 shows waveforms (including second order distortion waveforms)
at each stage of the circuit of Figure 2.
Figure 4 shows a graph of negative bias voltage vs. second order distortion
performance in dBc with optical input power of 1 mW.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT
The invention will first be described generally with reference to Figure 1.
The transimpedance amplifier 10 consists of four major components. First, an
optical input 15 is received into an optical to electrical transducer 20. The output of
the optical to electrical transducer 20 is a low voltage signal supplied into the input
stage 30 which serves to convert this low voltage signal to a current signal. The
voltage amplifier stage 40 receives the current signal, amplifies it, and feeds it into
an emitter follower stage 50. By design, the emitter follower stage 50 also creates
second order distortion products which are 180° out of phase with second order
distortion products created by the input stage 30. A feedback 60 flows from the
emitter follower stage 50 back to the input stage 30. The output signal 70 therefore
has reduced distortion levels because of the canceling effect of a controlled second
order distortion produced by the emitter follower stage 50.
An exemplary circuit 10 for achieving the invention of Figure 1 is shown in
Figure 2. This circuit 10 will now be described in greater detail. Exemplary
components and values are shown in Table 1. It should be understood by those
reasonably skilled in the art, however, that these components can be substituted and
the values may be modified to remain in conformance with the invention. An optical
signal 15 is supplied to photo diode PD. The photo diode PD is negative biased and
supplies an electrical signal to the input stage 30.
TABLE 1
PD Epitax EPM 650 R13 2K (1%)
Qi ATF10236 by HP R14 12K
Q2 BFT 92 or BFT 92w by Philips R15 12K
Q3 AT 41511 by HP R16 22
Rl 120 R17 12K
R2 22 Cl .luF
R3 2.4K C2 .luF
R4 IK C3 .luF
RS 22 C4 Λ F
R6 39 C5 ΛuF
R7 1.8K C6 ΛuF
R8 22 C7 6pF
R9 820 C8 ΛuF
RIO 750 C9 Λ F
Rll 68 CIO 18PF
R12 3.9K Cll ΛuF
The input stage 30 consists of transistor Ql3 resistors R,, R2 and R6 and
capacitors C2 and C4. The transistor Q, is preferably a GaAs MESFET. It should
be understood however that other active components exhibiting similar electrical
properties could be substituted. The gate of the transistor Q, is connected to the
photo diode PD. The drain of the transistor Q, is connected to the parallel
combination of resistor R2 and capacitor C2 while the source is connected to the
parallel combination of capacitor C4 and resistor R6.
The voltage amplifier stage 40 consists of transistor Q2, and resistors Rs, R7
and capacitor C10. The transistor Q2 is preferably a P-N-P BJT. It should be
understood, however, that other active components exhibiting similar electrical
properties could be substituted. The output of the transistor Qj goes to the emitter
of the transistor Q2 through resistor Rg and capacitor Cι0. The base of transistor Q2
is connected to a voltage source 35 through resistors R3, R4, R16 and capacitor C3.
Resistors R2, Rg, R16 and R8 are small value resistors used to prevent unwanted high
frequency oscillations.
The emitter follower stage 50 consists of transistor Q3, resistors R8, R9, Rn
and capacitor C8. The transistor Q3 is a N-P-N BJT. The collector of the transistor
Q2 is connected to the base of the transistor Q3 through resistor R8. The collector of
the transistor Q3 is connected to the voltage source 35. The emitter supplies a signal
to both the output 70 and the feedback loop 60. The feedback loop 60 consists of
parallel capacitor C7 and resistor Rι0 combination connected in series with resistor
R,2. Resistor R12 is connected to the gate of the transistor Qj. Resistor R7 is
connected between the collector of the transistor Q2 and a negative bias voltage
source 52. Resistor R<, is connected between emitter of Q3 and the negative bias
voltage source 52. The signal S3 from the emitter of transistor Q3 passes through
capacitor C8 and resistor Rn to output 70. An optional optical power monitor may
be located at either resistor R13 or Rπ.
Referring now to Figures 2 and 3, in operation, the photodiode PD supplies
an electrical signal S0 corresponding to the optical input 15 to the transistor Q, gate.
This signal S0 is usually a low voltage broadband radio frequency (RF) signal
containing many channels of data and is shown in Figure 3 as a representative
sinewave. The transistor Q, converts this voltage signal S0 to a current signal Ij
flowing in the first stage and second input stage. This current signal generates
voltage signal S} at the output stage of transistor Qv As shown in Figure 3, the
conversion through transistor Qj also introduces second order distortion signal Sld
causing greater amplification on the negative half cycle Sj. than on the positive half
cycle S1+. The current signal Ij is then amplified by the transistor Q2 in the output
stage of resistor R7. The output of the transistor Q2 collector is an amplified signal
S2.
The signal S2 is supplied to the transistor Q3 at its base. The transistor Q3 also
generates second order distortion products signal S3d which are inherently 180° out
of phase with distortion products in signal S2d. The amplitude of the distortion signal
S3d must be controlled to match the amplitude of distortion signal in S2d in order to
achieve distortion cancellation. Control of distortion signal S3d is achieved by first
determining the level of distortion signal S2d and then adjusting the negative bias
voltage source 52 and resistors R9 and Rπ to achieve a similar level distortion signal
S3d. Maximum distortion cancellation is achieved when the amplitude of signal S3d
is closely matched to the amplitude of signal S2d. The transistor Q3 also
characteristically amplifies the positive half cycle S3 + more than the negative half
cycle S3 " therefore correcting the amplitude variation in the half cycles of the signal
S2. The resultant output signal S3 as shown in Figure 3 is amplified, corresponds
with the input signal S0 and has minimal second order distortion.
Operation of the circuit shown in Figure 2, selection of components, and
distortion cancellation methodology can be further explained mathematically by first
estimating the distortion produced in the transistor Qt and then determining the
distortion of the transistor Q3. First, to estimate the distortion of transistor Qλ:
g„X- C ( l- ) (Eq. l)
C is a constant. Vgs is the voltage between gate and source. Vp is the
MESFET transistor Qj pinch-off voltage. gm is the transconductance of the
MESFET transistor Qj.
The output voltage of the emitter follower is basically equal to the output
voltage of the voltage amplifier stage 40. The output voltage of the voltage amplifier
stage 40 is given by:
V2 = - gMR,Vm (Eq. 2)
Suppose the DC bias voltage on transistor Qj gate is Vdc, and the RF signal
is Vιn.
The output voltage on the transistor Q2 collector will be:
According to the transimpedance amplifier, if the gain of transistor Q2 is large
enough, for the input optical power of Popt,ca, power, the output voltage of V3 will be:
V^ - PopticalPower R Rn OMI (Eq.4)
where R is the optical detector responsivity, OMI is the optical modulation
depth.
The voltage on the output transistor Q2 will be:
V
2 = - CR
7 1 - V... (Eq. 5)
Suppose the voltage amplifier stage gain is A,
According to Eq (3), the ratio of output power of the second order distortion
to the fundamental power will be:
A is the amplification of the voltage amplifier stage.
If Popticai power 1 mW, R is 0.8 ma/niW, OMI is 20%, R,2 is 5 k ohm. A is 120.
V p is 1.3V. Vdc is 0.6V. V3 is approximately 0.8 volts (peak value).
The second order distortion of first stage of GaAs MESFET transistor Q] will
be -41dB.
This means that the original PIN-GaAs MESFET transimpedance amplifier
will have a large second order distortion for lmW optical input power.
The second order distortion created by GaAs MESFET transistor Q, can be
canceled by the emitter follower stage 50 if the circuitry is designed correctly.
The emitter follower second order distortion comes from the internal emitter
resistance which is part of the emitter follower output impedance. This emitter
follower's internal resistance at room temperature can be calculated by:
26 R -*■ ^intern l resis tance = ~ IX (Eq *. 7) '
Where Ie is the emitter follower stage emitter current in milliamps. This
emitter current can be an instant current. When the emitter follower has lower output
RF impedance, the Ie swing can be large at large RF output. This Ie current swing
creates a nonlinear resistance in the output circuit, thus creates second order
distortion. Controlling the emitter follower DC part of Ie (by controlling the value
of R9 or the negative voltage power supply) and the RF output impedance, the output
of the second order distortion level can be easily controlled over a wide range. Thus
by correct design of the transimpedance circuitry, due to the cancellation of these
two distortions, very linear output can be realized.
For the total cancellation of the two-distortion signals S2d and S3d, the most
important conditions are that these to distortion signals must have the same
amplitude and be 180° out of phase.
For total cancellation, the same amplitude is guaranteed by the correct design
of the input stage 30 and adjustment of the parameters of the emitter follower stage
50. The correct phase relationship at low frequency is guaranteed by the circuit
topology.
The first stage MESFET transistor Qj is a current square device. In the first
stage input, the positive half cycle S,+ will get more amplification. At the MESFET
transistor Qx drain, the negative voltage half cycle gets more amplification. The
emitter follower transistor Q3 is the other second order distortion generator. At the
output terminal, positive voltage gets more amplification. Because the second stage
is a common base configuration, it has no phase inversion for the input and output.
So the two distortion signals are 180° out of phase at low frequencies.
For higher frequencies, the amplification of the voltage amplifier stage 40
becomes lower. This is because the capacitance in the output shunts the output.
This will create phase shift, thus reducing the cancellation of the second order
distortion. For transimpedance amplifiers, when input optical power is fixed, a
reduction in amplification of transistor Q2 will cause the input voltage of the input
stage 30 to become larger. Thus the GaAs MESFET transistor Q, will generate more
distortions. Because of this, the output signal will have larger distortion levels at
higher frequencies.
Figure 4 shows the experimental second order distortion data versus negative
power supply voltage measured at optical input power at 1 mW. For the circuit of
Figure 2, the minimum distortion is achieved at a negative bias voltage of -16V.
Utilizing the formulas and analysis those skilled in the art will appreciate that
variations to the circuit of Figure 2 are possible and distortion can be minimized if
the methods presented here are followed.